Gemini Omni

Google / Google DeepMind

Google's first unified multimodal model — "a model that can create anything from any input." Text, image, audio, and video go in; video comes out first, with image and audio output to follow. The framing collapses the toolchain that filmmakers had been stitching together — separate models for storyboard, motion, sound, dialogue — into a single context. The marquee demo: editing video conversationally, like talking to an editor.